Reducing Hallucination in Multimodal Large Language Models through Hard Grounding Preference Supervision
cs.CV, cs.AI, cs.CL, cs.DB, cs.LG
Submitted: 2026-05-13
Updated: 2026-09-26
License: http://creativecommons.org/licenses/by-nc-sa/4.0/
The gist: Hallucination remains a major challenge in vision-language models (VLMs), particularly when linguistically plausible responses are unsupported by visual evidence.
Terminology
Abstract
Hallucination remains a major challenge in vision-language models (VLMs), particularly when linguistically plausible responses are unsupported by visual evidence. We study whether multimodal hallucination can be reduced by concentrating post-training supervision on hard grounding boundaries, where preferred and rejected responses are semantically close but differ in their support from observable visual evidence. Under a frozen visual encoder and cross-modal alignment pathway, we first use supervised fine-tuning (SFT) to establish broad decoder-side multimodal behavior and then construct hard grounding preference pairs for Direct Preference Optimization (DPO). These pairs target evidence utilization, calibration, and grounding consistency across fine-grained recognition, spatial reasoning, OCR, ambiguous or insufficient evidence, and false-premise queries. The resulting DPO model consistently improves over its SFT initialization on DocVQA, TextVQA, MMBench, and VQAv2. Across approximately 839 additional multimodal examples, it further achieves 6.2--8.2 percentage-point higher pairwise win rates under three independent LLM judges. We additionally release HardVQA-DPO, a curated and growing resource with more than 3K hard-grounding SFT examples and an initial set of DPO preference pairs.https://huggingface.co/datasets/vlmgrounding/hardvqa-sft-dpo These results show that decoder-side adaptation alone can improve a meaningful subset of grounding failures under a fixed multimodal representation, while preserving broader multimodal capabilities.
Sources
- MMBench: Is Your Multi-modal Model an All-around Player?
- Self-contradictory Hallucinations of Large Language Models: Evaluation, Detection and Mitigation
- Multilingual OCR-Aware Fine-Tuning and Prompt-Guided Chain-of-Thought Reasoning for Multimodal Large Language Models
- Robust Checkpoint Selection for Multimodal LLMs via Agentic Evaluation and Stability-Aware Ranking
Related papers
- Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
- AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
- Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
- MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
- TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
- A Survey on Efficient Vision-Language-Action Models